Returning 2 distinct columns from a 3-column result-set

MaxOrDefaultPhoneNumber.sql

-- We want to return one phone number per each person-id
-- If a phone number is flagged as default it has a higher priority
-- There may be more than one phone numbers be marked as default, 
-- or one person may have several phone numbers none of which is marked as default


-- Table Definition
CREATE TABLE [dbo].[PersonPhone](
	[PersonId] [int] NOT NULL,
	[Phone] [varchar](50) NOT NULL,
	[IsDefault] [bit] NOT NULL
) 

GO

-- Tab-separated sample data 
--
--PersonId	Phone	IsDefault
--1	123	True
--1	234	False
--1	324	False
--2	1234	False
--2	23234	False
--3	23948	True
--3	213432	False
--3	9328	False
--4	29343	False
--4	492384	False
--5	93432	True
--5	3984	True
--5	821433	False
--6	23943	False
--6	237423	False
--6	324234	True
--6	233434	False

-- solution 1

SELECT Ph.PersonId, 
(
    SELECT TOP 1 Ph2.Phone FROM PersonPhone Ph2 
    WHERE Ph2.PersonId = Ph.PersonId
    ORDER BY Ph2.IsDefault DESC
) AS Phone

FROM PersonPhone Ph
GROUP BY Ph.PersonId 

GO

-- Solution 2

SELECT Ph.PersonId, Ph.IsDefault, MAX(Ph.Phone)
FROM PersonPhone Ph
INNER JOIN 
(
    SELECT Ph2.PersonId, MAX(CONVERT(INT, Ph2.IsDefault)) AS MaxDef 
    FROM PersonPhone Ph2 
    GROUP BY Ph2.PersonId
) T 
ON Ph.PersonId = T.PersonId AND Ph.IsDefault = T.MaxDef
GROUP BY Ph.PersonId, Ph.IsDefault

GO

-- TODO: Which of them is better and why?
-- Turning on Client Statistics on SQL Server Express 2012 on both above queries, 
-- reveal that the first solution is less efficient on processing and waiting time 
-- compared to the second one. Conversely, the first query is more efficient in 
-- terms of the number of bytes received and sent.
-- However the first solution is much easier to write for a programmer. Imagine for
-- a second that what would happen if we wanted to have 3 or 4 distinct columns in
-- a result set and how would that be implemented using the second query.


-- A definitely WRONG answer

SELECT Ph.PersonId, MAX(CONVERT(INT, Ph.IsDefault)), MAX(Ph.Phone)
FROM PersonPhone Ph
GROUP BY Ph.PersonId

GO